<img height="1" width="1" style="display:none;" alt="" src="https://dc.ads.linkedin.com/collect/?pid=358169&amp;fmt=gif">

Is the Balance Between Art and Science Switching in Forecasting?

Machine-Learning

Forecasting experts have traditionally operated within the grey area between science and art. While they use hard data, statistics, and other scientifically-backed techniques to gather information and build a strong statistical baseline forecast, they ultimately rely on their own expertise when adjusting calculations and drawing conclusions. However, the trend towards automation and the rise of big data, machine learning, and other AI-powered technologies has led many to wonder whether the scales are tipping. Can the art of interpreting the data and making strategic business decisions — once the sole domain of demand planners — be offloaded to advanced algorithms?

 

Who makes better predictions: humans or machines?

As humans, our biases exert a strong influence over what we believe will happen in the future. In fact, it has long been established that, when compared with human judgment, even simple algorithms can produce more accurate predictions. Be that as it may, data scientists still leaned heavily on input from industry experts when designing these algorithms in order to decide which factors should be measured and what weight those factors should be given.

Yet this has been changing. In 2015, researchers at MIT created a machine equipped with AI technology that could analyze patterns and identify the most relevant data sets before making a prediction. And it performed remarkably well. When pitted against humans in three separate competitions, it produced more accurate predictions than two-thirds of the participating teams — and it did so in a fraction of the time. Given the speed with which technology is advancing, it’s not implausible that these AI-powered machines will become even better over the next few years at identifying which factors are relevant to forecasting and then weighting them appropriately.

 

The limitations of technology

So if machines are better at making predictions, where do humans fit in? Do demand planners still add value to an organization’s supply chain management strategy? The answer is undoubtedly yes. While it is true that technology has enabled us to make better predictions, ones that aren’t as subjective or informed by our internal biases, there are limitations to what can be automated.

In an interview with Harvard Business Review, Avi Goldfarb, co-author of the book Prediction Machines: The Simple Economics of Artificial Intelligence, asserts that where humans excel — and where machines fall short — is in deciding what predictions to make and how to apply the knowledge gained. Demand planners and sales forecasters, for example, have a wealth of knowledge about their company and their chosen industry. They understand their company’s mission, values, and overall growth strategy better than any machine. And as a result, they are much better equipped to make strategic, growth-oriented business decisions. In short, machines may provide the prediction, but humans provide the context and manage its application.

 

Demand planning in the 21st century

There is a sweet spot, then, between an entirely manual process and full automation. Forecasting is both a science and an art. To be successful, companies must understand the value of AI and machine learning and use those technologies to augment the role of a demand planner, but also recognize that a human touch is still necessary for long-term growth.

This is certainly what we’ve seen with our clients. Milwaukee Tool, for example, has used our Forecastability Analysis service to help determine what data their employees should collect and how the company as a whole should be structured to better support its forecasting initiatives. And Deschutes Brewery has employed our Atlas Planning Suite to predict customer demand for existing and new products with greater precision, avoid stock-outs, and minimize overstocking. Because of our partnership, the brewery’s supply chain managers and demand planners are able to spend less time and energy on data collection and analysis and more on the efficient application of that information.

However we’ve also noticed that companies aren’t always fully sure where to start. However many years of experience in supply chain management you have, you may be a little hesitant when it comes to navigating the grey area between the art and the science of forecasting. This is where having a strong partner who understands your business goals, can identify appropriate software solutions, and help you through the integration process can be extremely valuable. Schedule your free free demo with John Galt’s forecasting experts now and learn how we can help you strike that balance and set you up for success.

 




Tags:     For Supply Chain Nerds How-to

Subscribe